4

Three Old Indian Values of π

Abstract
Three values of π from the Śatapatha Brāhman.a
and Baudhāyana Śulbasūtra are discussed. These val-
ues emerge when squares are transformed into cir-
cles of equal area, a commonly occurring operation
in Vedic altar construction.
KEYWORDS: Ancient geometry, π

1 Introduction
In this article we describe three hitherto neglected references
to π, one from Śatapatha Brāhman.a (ŚB) and the others
from Baudhāyana Śulbasūtra (BŚS). These references relate
to construction of altars of certain shapes and sizes the back-
ground to which is described in the analysis of Vedic geometry
by Seidenberg1 . Histories of Indian mathematics generally
begin with the geometry of the Śulbasūtras but Seidenberg
showed that the essentials of this geometry were contained in
the altar constructions described in the much older Śatapatha
Brāhman.a and Taittirı̄ya Sam . hitā. More recently it has been
shown that Vedic astronomy goes back to at least the third
millennium B.C.2 and so a concomitant mathematics and ge-
ometry must have existed then. It has also been shown that
an astronomy is coded in the organization of the Vedic books
itself.3
Meanwhile archaeological discoveries have had major im-
plications for the understanding of ancient Indian chronology.4 Briefly,
discoveries related to the drying up around 1900 B.C. of the
Sarasvati river, the preeminent river of the Rigvedic age, in-
dicate that this epoch must be considered the final limiting
point in time for the Rigveda. Therefore, it seems reason-
able to assign second millennium B.C. for the Brāhman.a
literature as is also attested by their internal astronomical
evidence.5 The Śulbasūtras have been traditionally dated to
later than 800 B.C., but Seidenberg suggests that their knowl-
edge belongs to a much earlier period.
Elsewhere, I have argued for a value of π implicit in the
organization of the Rigveda and this should be earlier than
the age of the Brāhman.a literature. But here we are con-
cerned only with early explicit values of π. The ŚB is not
mentioned as a text in the list of values of π in the review
by Hayashi et al6 , who have missed the antecedents of the
history of π in India. We also show a connection between the
approximations of ŚB and BŚS.

2 Transforming a square into a circle

The tranformation of a square altar into a circular one is a
commonly occurring theme in the altar ritual. This leads to
many geometric and algebraic results and also many values
for π.
ŚB 7.1.1.18-31 describes the construction of a circular
gārhapatya altar starting with bricks of different kinds.

sa catasrah. prācı̄rūpadadhāti dve paścāt tiraścyau

dve purastāt. tad yāścatasrah. prācı̄rūpadadhāti
sa ātmā. tad yat tāścatasro bhavanti caturvidho
hyayamātmā ’tha ye paścāt te sakthyau ye purastāt
tau bāhū yatra vā ’ātmā tadeva śirah.. [ŚB 7.1.1.18]
He puts on (the circular site) four (bricks) running
eastwards; two behind running crosswise (from
south to north), and two (such) in front. Now
the four which he puts on running eastwards are
the body; and as to there being four of these, it is
because this body (of ours) consists of four parts.

30
 The two at the back then are the thighs; and the
two in front the arms; and where the body is that
(includes) the head. [Eggeling’s translation]

A figure detailing this construction (part 3, page 302) is

given by Eggeling.7 Elsewhere, as in ŚB 10.2.3.1-3, there is
unmistakable mention of an āhavanı̄ya altar of one vyāma
square. The relevant passage from ŚB 10.2.3.3 is:

[The space of a] vyāma which was (marked off),

is the womb of the gārhapatya, for it was from
that womb that the gods begat the gārhapatya;
and from the gārhapatya the āhavanı̄ya.

Later texts such as the Śulbasūtras explicitly speak of

the gārhapatya altar being equal to one vyāyāma measure,
in either the square or the circular form (see e.g. BŚS 7.4-5).
Seidenberg analyzed these constructions in his papers and
he concluded that the gārhapatya and the āhavanı̄ya altars,
generally circular and square respectively, were of equal area.
But he added that “one will not come, without interpretation,
to an unambiguous meaning from such passages.”8
For ready reference we provide the relevant units of mea-
surement:

1 purus.a = 1 vyāma = 5 aratnis = 120 aṅgulas

We argue here that ŚB, in fact, does provide conclusive ev-

idence of the identity of the areas of the two altars. The con-
struction of the gārhapatya altar is using oblong and square
bricks (Figure 1). The bricks come in a variety of sizes and
the question is to determine whether the context can fix the
specific size meant in ŚB.
The square bricks used are either one-fourth, one-fifth,
one-sixth, or one-tenth of a purus.a on each side.9 From the
nature of the construction the only square brick of the types
that will fit into the scheme is the one-fifth, the pañcami,
which is 24 × 24 square aṅgulas which is the same as 1 × 1

31
 Figure 1: Identity of circular and square altars

square aratni. The other bricks lead to area that is either too
large or much too small to fit our equal area requirement. The
oblong brick is therefore 48 × 24 square aṅgulas.
Since the inscribed square in Figure 1 is 45 × 45 square
purus.a, √
we see that the diameter of the circular gārhapatya
altar is 32/5 purus.a. Since its area is taken to be equal to
that of the one square purus.a āhavanı̄ya altar, this leads to

π × 32/25
= 1. (1)
4
From this it follows that
25
π1 = . (2)
8

3 A construction from BŚS

Sen and Bag10 have described several approximations of π in
the Śulbasūtras. But here we are interested in BŚS 16.6-

32
11, which gives a construction for converting a square into
a chariot-wheel or a wheel with spokes (rathacakra). As we
will see this starts with a very specific design using bricks of
a certain size, but ultimately the design is described in bricks
of entirely different shapes with circular arcs.
In this design one starts with an area equal to 225 square
bricks; this is now augmented by 64 more bricks so that one
has a new square equal to 172 = 289 bricks. Each brick is to
1
be of an area 30 square purus.a.

tāsām
. dve śate pañcavim . śatiśca sāratniprādeśah.
saptavidhah. sam . padyate (16.8)
tāsvanyāścatuh. s.as..timāvapet. tābhih. samacaturaśram .
karoti.
tasya s.od.aśes..takā pārśvamānı̄ bhavati.
trayastrim . śadatiśis. yante tābhirantānsarvaśah. paricinuyāt.
(16.9)
nābhih. s.od.aśa madhyamāh..
catuh.s.as..tirarāścatuh. s.as..tirvedih.. nemih. śes.āh. (16.10)

With 225 of them [bricks] is produced the seven-

fold [altar] with two aratnis and [one] prādeśa.
(16.8)
To these [225] another 64 [bricks] are added and
with them a square is made. (At first) a square
is made with a side containing 16 bricks, leaving
a balance of 33 bricks. These are placed on all
sides.(16.9)
16 (bricks) at the centre constitute the nave; 64
(bricks, thereafter) constitute the spokes and 64
the empty spaces (between the spokes); the re-
maining (bricks) form the felly. (16.10)

The reference in the second part of 16.8 is to the area of

the altar which is to be 7 12 square purus.a as 2 aratnis and
one prādeśa equal half a purus.a. The main fire altars in the
agnicayana ritual were to be of this size.

33
 15 17

19
Figure 2: Circular area of diameter 17 units

The construction that follows begins with a square of 225

units which is later enhanced to a square of 256 units. The
remaining 33 units are distributed around this larger square
to give us a circular altar of diameter 17. Sen and Bag have
taken this distribution of 33 units to lead to another square
of 17 × 17 and that is correct as far as the design goes, but we
will show later that there is also the intent make a circular
area of diameter 17 units (Figure 2).
The next sūtra speaks of the further development of the
design into a wheel with spokes: the centre (nave), the middle
(spokes and spaces), and the outer (felly). Next, BŚS 16.11
speaks of nemimantataścāntarataśca parilikhya, or the “outer
and the inner enclosing the felly done into circles”, but there
is no reason to assume that this is not a reiteration of the
step carried out in BŚS 16.9.

34
 If we accept for the moment that we have an equality of
a square of side 15 units into a circle of diameter 17 units,
289
π× = 225. (3)
4
Or, we obtain the approximation
900
π2 = . (4)
289
The important point in this transformation of the square
into the circle is that the diameter of the circle is two units
greater than the side of the square.
Note that the inner square of 16 bricks becomes the nave
and, of the next set of 128 bricks, exactly half go in to the
formation of the spaces between the spokes. This means that
64 bricks are left over; these are removed and so the area of
the altar remains 7 12 square purus.a.
One might object that the tranformation of the outer
square into the circular felly with diameter of 19 units means
that there was no need to convert the first square of 15 × 15
into the circle of diameter 17 units. But, in reality, we do
need a circle in the middle of the felly; this is described in
the sūtra 16.12:

nemim. catuh.s.as..tim
. kr.tvā vyavalikhya madhye parikr.s.et.
(16.12)

After dividing the felly into 64 equal parts and

drawing lines, a circle is drawn through the mid-
dle (of the felly). (16.12)

Although this middle is not specified, it is almost certain

that it is the first circle of diameter 17. That is because
the separation between these two circles is one brick-width
on each side. This strengthens the view that the sūtra 16.9
establishes the method which is used in going from the square
of 17×17 to a circle of diameter 19. This construction requires
first increasing two sides of the original square by one unit,

35
and then distributing the remaining bricks to round off the
sides as in Figure 2.
The outer construction corresponds to a value of π equal
to
361
π× = 289. (5)
4
Or,
1156
π3 = . (6)
361
So we have two further values for π, and it is interesting to
note that one of these is very similar to the one in ŚB which
can be expressed as 900 288 . It is plausible that the approxima-
tion (4) arose from the realization that 1 more than 288 was
a square. But the approximation (4) is worse than that of
(2). A return to the earlier value is seen in the later Mānava
Śulbasūtra (MŚS) where the chariot-wheel (rathacakra) altar
uses 344 bricks instead of the 289 of BŚS. The details of the
construction are not clear but it appears that the value of
π was 1075344 . Support for such an interpretation comes from
the rule MŚS 11.15 where we encounter precisely the same
value.11 Furthermore, since these texts list a variety of values
which, it is clear, were taken to be approximations.
The conversion of squares into circles is the basic geomet-
ric issue in BŚS 16, which can be seen from the nature of the
Figure 3 emerging at the conclusion of the constructions.12 Notice
that the original bricks are now replaced by new kinds of
“bricks” which total 200 as required by the rules of altar de-
sign. These new bricks are segments obtained by dividing a
circle radially and circularly.
It is significant that the representation of a square of side
15 by the circle of diameter 17, and that of a square of side
17 by the circle of diameter 19, are two of the best three such
approximations that can be obtained where the difference in
each set of numbers is 2. An even better approximation is to
represent a square of side 16 by the circle of diameter 18; but
these numbers do not arise in this altar design.

36
 Figure 3: A circular altar

4 Concluding Remarks
The altar constructions in the Śulbasūtras use bricks of such
great variety that it is likely that in most cases the “ritual”
represented just conceptual geometric exercises. This may be
seen in Figure 3 which gives the final arrangement of bricks
in the first layer of the chariot-wheel altar. This first layer
requires seven different types of bricks and the next layer
requires another nine different types. Most of these bricks
are not rectangular. It is highly unlikely such bricks were
actually made using casts. Ancient ruins show no trace of
these bricks.
Śulbasūtra constructions may be seen as defining a dis-
crete geometry. The use of “bricks” implies changes by pre-
defined elementary areas.
This note supports the view that the Śulbasūtras repre-
sent a continuation of the geometric tradition of the Brahman.as.
Furthermore, the relationship between the chariot-wheel con-

37
structions of two different texts indicates that many different
approximations for π were used.

References and Notes

1. A. Seidenberg, “The origin of mathematics,” Archive
for History of Exact Sciences, 18 (1978), 301-342; A.
Seidenberg, “The geometry of the Vedic rituals,” In
Agni, F. Staal (ed.), Berkeley, 1983.

2. D. Frawley, “Planets in the Vedic literature,” Indian

Journal of History of Science, 29 (1994), 495-506; S.C.
Kak, “Knowledge of the planets in the third millennium
BC,” Quarterly Journal of the Royal Astronomical So-
ciety, 37 (1996), 709-715.

3. S.C. Kak, The Astronomical Code of the R. gveda. Aditya,

New Delhi, 1994; see also S.C. Kak, “The astronomy of
the age of geometric altars,” Quarterly Journal of the
Royal Astronomical Society, 36 (1995), 385-396.

4. G. Feuerstein, S.C. Kak, D. Frawley, In Search of the

Cradle of Civilization. Quest Books, Wheaton, 1995.

5. P.C. Sengupta, Ancient Indian Chronology. University

of Calcutta, Calcutta, 1947.

6. T. Hayashi, T. Kusuba, and M. Yano, “Indian values

of π derived from Āryabhat.a’s value,” Historia Scien-
tiarum, 37 (1989), 1-16.

7. J. Eggeling, (tr.), The Śatapatha Brāhman. a. Motilal

Banarsidass, Delhi, 1988.

8. A. Seidenberg, “The geometry of the Vedic rituals,” In

Agni, F. Staal (ed.), Berkeley, 1983, page 114.

9. S.N. Sen and A.K. Bag, The Śulbasūtras. Indian Na-

tional Science Academy, New Delhi, 1983.

38
10. S.N. Sen and A.K. Bag, The Śulbasūtras.

11. This figure, taken from the book by Sen and Bag (page
218), does not show the circle dividing the felly accord-
ing to the interpretation in this paper.

12. R.C. Gupta, “New Indian values of π from the Mānava

Śulba Sūtra,” Centaurus, 31 (1988), 114-126.

39